JP2002018274A - Method for operating treatment apparatus and method for detecting abnormality of treatment apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve such a problem that there is no technique objectively judging a high-frequency power supply or a treatment chamber of a treatment apparatus is stabilized heretofore, and the experience and perception of an operator is required in order to judge whether the high-frequency power or the treatment chamber of the treatment apparatus is stabilized. SOLUTION: A method for estimating the attrition rate of expendables has a process for preliminarily calculating a standard residual score Q0 using the measuring data of the high-frequency power supply 18 of the stabilized treatment apparatus 10, a process for measuring a plurality of electrical data of the arbitrary treatment apparatus 10 immediately after starting, a process for calculating a comparing residual score Q using the high-frequency power supply and a precess for comparing both Q, Q0 to detect the stable state of the high-frequency power supply 18 of the arbitrary treatment apparatus 10 from the difference (Q-Q0) of both of them.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for operating a processing apparatus and a method for detecting an abnormality in the processing apparatus. The present invention relates to a method of operating a processing device that can be operated in a state where the temperature has reached the maximum value, and a method of detecting abnormality of the processing device.

[0002]

2. Description of the Related Art A processing apparatus (hereinafter, simply referred to as a "processing apparatus") is used for an etching process, a film forming process and the like. This type of processing apparatus applies, for example, high-frequency power to electrodes in a processing chamber and introduces a process gas into the processing chamber.
A plasma of a process gas is generated in the processing chamber, and a target object such as a semiconductor wafer is subjected to a predetermined plasma process. Although a high-frequency power supply is used in the processing apparatus, the processing of the object to be processed is performed after the high-frequency power supply is stabilized according to the state in the processing chamber. However, immediately after the start of the processing apparatus, the high-frequency power supply is unstable and not stable for a long time until it adapts to the state of the processing chamber.

[0003] For example, FIG. 8A is a diagram showing a change in a parameter (voltage) related to a high frequency of a matching circuit, and FIG. 8B is a diagram showing a parameter (electricity) of a capacitor characterizing the matching state of the matching circuit. FIG. 5 shows the fluctuation of the capacitance, but it is difficult to judge the stable state because of the fluctuation. With the parameters shown in FIG. 8A, a peak at the beginning of the lot is observed, but it is difficult to determine whether or not the stabilization has been achieved. In addition, the processing chamber requires a considerable amount of time to adapt to the environment to which high-frequency power is applied, and is not easily stabilized. For this reason, conventionally, it is determined whether or not the high-frequency power supply and the processing chamber are stabilized based on the experience and intuition of the operator, and when it is determined that the stable area has been reached, an object to be processed such as a wafer is put into the processing chamber and the predetermined processing is performed. Was given. The operating conditions in FIG.
After the maintenance and inspection, the processing chamber is evacuated for 4 days, and the condition of little deposition is shown. The conditions for less deposition will be described later.

[0004] Also, when maintenance and inspection of the processing apparatus are performed, consumables are replaced and cleaning is performed. However, since the processing apparatus is a precision machine, careful assembly is required.
For example, the plasma becomes unstable if there is any looseness in the screwing of the high-frequency power supply or each component in the processing chamber, or if there is a mistake in mounting the components. Therefore, such mistakes are far from tolerable.

[0005]

However, conventionally, there is no method of objectively determining whether the high-frequency power supply of the processing apparatus or the processing chamber is stable or not, and the operator must rely on the experience and intuition of the operator. Establishment of the objective method is expected. Further, it is not possible to evaluate processing conditions for bringing the processing apparatus to a stable state, and thus the evaluation has to rely on trial and error.

Conventionally, if a processing apparatus is operated without noticing a component mounting error, there is no method for detecting the mounting error without opening the processing apparatus. Needed.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide a method of operating a processing apparatus capable of objectively judging a stable state of the processing apparatus after starting and optimizing a condition. It is intended to provide. It is another object of the present invention to provide a method of detecting an abnormality of a processing apparatus, which can reliably detect an abnormality due to a component mounting error without opening the processing apparatus, and can classify the mounting error.

[0008]

According to a first aspect of the present invention, there is provided a method of operating a processing apparatus, wherein a high-frequency power is applied to an electrode in a processing chamber from a high-frequency power source to generate plasma, thereby processing an object to be processed. At this time, a multivariate analysis is performed using a plurality of measurement data measured by a measuring instrument that measures a plurality of electrical data of the high-frequency power supply that changes according to a state in the processing chamber, and an application state of the high-frequency power supply Is a method of operating the processing apparatus by detecting the multivariate analysis for the reference in advance using the measurement data when the application state of the high-frequency power supply is stabilized according to the state of the processing chamber of the processing apparatus. Performing, a step of measuring a plurality of electrical data of any processing apparatus immediately after starting, a step of performing a multivariate analysis for comparison using a plurality of measured data, a result of a comparative multivariate analysis and a reference multivariate. Compare analysis results The high frequency power from the difference between both the arbitrary processor Te is characterized in that a step of detecting that has reached a steady state according to the state of the processing chamber.

In the method of operating a processing apparatus according to a second aspect of the present invention, when a high-frequency power is applied to an electrode in a processing chamber from a high-frequency power source to generate plasma and process a target object, Detecting the application state of the high-frequency power supply by performing a principal component analysis using a plurality of measurement data measured by a measuring instrument that measures a plurality of electrical data of the high-frequency power supply that changes according to the state in the processing chamber A method of operating the processing device, a step of performing a principal component analysis for reference in advance using the measurement data when the application state of the high-frequency power supply is stabilized according to the state of the processing chamber of the processing device, Immediately after starting, measuring a plurality of electrical data of any processing device, performing a principal component analysis for comparison using multiple measured data, comparing the results of the comparative principal component analysis and the results of the reference principal component analysis And then both From one in which the high-frequency power supply of the arbitrary processing apparatus, comprising the step of detecting that has reached a steady state according to the state of the processing chamber.

According to a third aspect of the present invention, there is provided a method of operating a processing apparatus according to the first or second aspect, wherein the electrical data includes at least a voltage value of each of a fundamental wave and a harmonic, It is characterized by using a current value, an impedance and a phase angle.

According to a fourth aspect of the present invention, there is provided a method for operating a processing apparatus according to the second or third aspect, wherein the residual score or the principal component score is obtained by principal component analysis. It is characterized by comparing the residual score or the principal component score.

According to a fifth aspect of the present invention, there is provided a method for operating a processing apparatus according to any one of the second to fourth aspects, wherein the processing conditions are based on a comparison result of residual scores. And / or determining operating conditions.

In the method for detecting an abnormality of a processing apparatus according to claim 6 of the present invention, when a high-frequency power is applied to an electrode in a processing chamber from a high-frequency power source to generate plasma and process an object to be processed, Detecting the application state of the high-frequency power supply by performing a multivariate analysis using a plurality of measurement data measured by a measuring device that measures a plurality of electrical data of the high-frequency power supply that changes according to the state in the processing chamber A method for detecting abnormalities in the processing apparatus by using the measurement data when the application state of the high-frequency power supply is stabilized according to the state of the processing chamber of the normal processing apparatus. Performing a multi-variate analysis using a plurality of measurement data, performing a multi-variate analysis for comparison using a plurality of measurement data, a comparison multi-variate analysis result and a reference multivariate analysis result. Compare both From difference is characterized in that a step of detecting an abnormality of said any processor.

According to a seventh aspect of the present invention, there is provided a method for detecting abnormality in a processing apparatus, comprising the steps of: applying a high-frequency power from a high-frequency power source to an electrode in a processing chamber to generate plasma; Detecting the applied state of the high-frequency power supply by performing principal component analysis using a plurality of measurement data measured by a measuring device that measures a plurality of electrical data of the high-frequency power supply that changes according to the state in the processing chamber A method for detecting an abnormality of the processing apparatus by using the measurement data when the application state of the high-frequency power supply is stabilized according to the state of the processing chamber of the normal processing apparatus. Performing a step of measuring a plurality of electrical data of an arbitrary processing device; a step of performing a principal component analysis for comparison using the plurality of measurement data; a result of the comparative principal component analysis and a result of the reference principal component analysis Compare both From difference is characterized in that a step of detecting an abnormality of said any processor.

In the method for detecting an abnormality of a processing apparatus according to claim 8 of the present invention, in the invention according to claim 6 or 7, at least a voltage value of each of a fundamental wave and a harmonic is used as the electrical data. , Current value, impedance and phase angle.

According to a ninth aspect of the present invention, there is provided a method for detecting an abnormality in a processing apparatus according to the seventh or eighth aspect, wherein a residual score or a principal component score is obtained by principal component analysis. , And comparing the residual score or the principal component score.

Further, according to a tenth aspect of the present invention, there is provided a method for detecting an abnormality in a processing apparatus according to any one of the fifth to ninth aspects.
In the invention described in the section, an abnormal part is classified according to a component of a residual matrix.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below based on the embodiments shown in FIGS. First, an example of a processing apparatus to which a method of operating a processing apparatus and a method of detecting an abnormality of the processing apparatus according to the present invention will be described with reference to FIG. As shown in FIG. 1, for example, a processing apparatus 10 used in the present embodiment includes a grounded processing chamber 11 made of a conductive material such as aluminum, An electrode 12 also serving as a mounting table for mounting a wafer W thereon, and a magnetic field forming means 13 for applying a rotating magnetic field, and an electric field generated between the upper and lower electrodes of the processing chamber 11 under the control of the controller 14. The rotating magnetic field B generated by the magnetic field forming means 13 acts to perform uniform plasma processing on the wafer W with high-density plasma. A gas supply pipe 15 is connected to the processing chamber 11, and supplies a process gas into the processing chamber 11 from a gas supply source (not shown) via the gas supply pipe 15. A gas exhaust pipe 16 connected to a vacuum exhaust device (not shown) is connected to a side surface of the processing chamber 11, and the inside of the processing chamber 11 is depressurized and maintained at a predetermined degree of vacuum through the vacuum exhaust device and the gas exhaust pipe 16. . A high-frequency power supply 19 is connected to the electrode 12 via a harmonic measuring device 17 and a matching circuit 18. High-frequency power is applied from the high-frequency power supply 19 to the electrode 12 to generate a process gas plasma in the processing chamber 11. For example, a predetermined etching process is performed on the surface of the upper semiconductor wafer W. Further, a focus ring 20 is provided on the periphery of the electrode 12.
Are arranged, and the plasma is focused on the wafer W via the focus ring 20.

In this embodiment, 13.56 MH is used.
z is applied to the electrode 12,
In addition to the 13.56 MHz high-frequency power, harmonics having this as a fundamental wave (for example, 27.12 MHz, 40.68 MHz)
MHz). However, electrical data such as voltage, current, phase, and impedance are unstable immediately after the start of the processing device 10 and are not easily stabilized. Moreover, there is no way to objectively know the state inside the processing chamber 11. Therefore, in the present embodiment, the electrical data such as the voltage, current, phase, and impedance are measured, and the measured values are used to stabilize the processing apparatus 10, specifically, a predetermined state in the processing chamber 11. A stable state required for plasma processing is detected.

That is, the fundamental wave of the high frequency power supply 19 and the voltage, current, phase and impedance of the harmonic wave are measured using the harmonic measuring device 17 interposed between the electrode 12 and the matching circuit 18 from the start of the processing device to the high frequency. The measurement is intermittently performed until the power supply 19 is stabilized, and these measured values are sequentially taken into the control device 14. The control device 14 stores, for example, a program for principal component analysis as a multivariate analysis program, and performs a principal component analysis of measured values via this program to detect a stable state of the processing device.

For example, when principal component analysis is performed in the present embodiment, a processing device in which the state of application of the high frequency power supply 19 to the electrode 12 in the processing chamber 11 is stabilized (hereinafter referred to as a “reference processing device”). ), The voltage, current, phase and impedance of the fundamental wave and its harmonics of the high-frequency power supply 19 are intermittently measured as electrical data, and the measured values V (f _n ) and I (f _n ) of each frequency are measured. , P (f _n ) and Z (f _n ).
Then, the relative values of these various measured values are obtained, and the various measured values are rendered dimensionless. At this time, the correspondence between various measured values and the dimensionless relative values is clarified, for example, according to the arrangement order of the measured data. Next, various dimensionless measured values (hereinafter, simply referred to as “relative measured values”).
When the number of measurements is n and the measurement is performed m times until it becomes stable, a matrix containing all the relative measurement data of the reference processing device is expressed by Expression 1. Next, the controller 14 calculates an average value, a maximum value, a minimum value, and a variance value based on all the relative measurement values, and then uses a variance-covariance matrix based on these calculated values to obtain a plurality of dimensionless measurement data. Is performed to obtain an eigenvalue and its eigenvector. The eigenvalue indicates the magnitude of the variance of the dimensionless measurement data, and is defined as a first principal component, a second principal component,..., An nth principal component in the order of the magnitude of the eigenvalue. Also, each eigenvalue has an eigenvector belonging to it. Normally, the higher the degree of the principal component, the lower the contribution to the evaluation of the data, and the less its usefulness.

(Equation 1)

For example, n relative measurement values are taken in each of the m measurements, and the j-th principal component corresponding to the j-th eigenvalue of the i-th measurement is expressed by Equation 2. Then, the i-th relative measurement value (x _i1 ,
x _i2 ,..., x _in ) are the scores of the j-th principal component in the i-th measurement. Accordingly, the score t _j of the j-th principal component is defined by Expression 3, and the eigenvector P _j of the j-th principal component is defined by Expression 4. t _j is a score representing the relationship between the measurements. Further, the eigenvector P _j is representative of the weight between measurements. Then, the score t _{j of} the j-th principal component
Represented by using the number 5 of the matrix X and eigenvectors P _j. The matrix X is expressed by Equation 6 using the scores of the principal components and the respective eigenvectors.

(Equation 2)

(Equation 3)

(Equation 4)

(Equation 5)

(Equation 6)

Therefore, in the principal component analysis, even if there are many types of measurement data, for example, the first and second principal components and at most the third principal component are collected as a small number of statistical data, and the small number of statistical data are collected. The driving condition can be evaluated and grasped simply by checking. For example, in general, if the cumulative contribution ratio of the eigenvalues of the first and second principal components exceeds 90%, the evaluation based on the first and second principal components becomes highly reliable. The first principal component indicates the direction in which the measured data is most dispersed as described above, and serves as an index for comprehensively evaluating the operating state of the processing apparatus, and is suitable for determining and evaluating the change over time in the operating state of the processing apparatus. ing. The second principal component is dispersed in a direction orthogonal to the first principal component and serves as an index of an instantaneous deviation from a normal operation state, and is suitable for judging and evaluating a sudden change in the operation state.

However, the first principal component can be generally evaluated by observing eigenvectors, the first principal component score, and the like to see how the data tends to be. Since each eigenvector is uniquely determined, it is not possible to grasp from various aspects how the individual measurement data is in each state and how it changes. Therefore, in the present embodiment, as a method of detecting that the application state of the high-frequency power supply 49 has reached a stable state according to the inside of the processing chamber 11, the (k + 1) th low-contribution ratio is used.
The residual matrix E defined by Equation 7 in which the above high-order principal components are integrated into one (the components in each row correspond to the relative measurement values of the high-frequency fundamental wave and its harmonics, and the components in each column are measured. (Corresponding to the number of times). Then, when this residual matrix E is applied to Expression 6, Expression 6 is expressed by Expression 8. Moreover, it obtains a residual score of residual matrix E of the reference processing unit based residual score Q _0, any processor to the residual score Q ₀ to the reference (hereinafter referred to as "comparison unit".) Is detected until it reaches a stable state after starting. In general, the residual score Q _i
Is expressed as the product of the row vector e _i and its transposed vector e _i ^T , and is the sum of the squares of the respective residual components, so that the residual components can be reliably obtained without canceling the plus and minus components. . Therefore, comparison processing unit can determine whether the reached steady state by comparing the residual scores Q _i of the comparison processing unit and residual scores Q ₀ of the reference processing unit for each measurement. Then, row vector of each row represented by the number 10 of the residual matrix E if residual score Q _i at point in the comparison processing unit is out of the residual scores Q ₀ of the reference processing unit at the same time Looking at the components of e _i, can there was a significant deviation in any of the measured values at that time understand, identify the cause of the abnormality.

(Equation 7)

(Equation 8)

(Equation 9)

(Equation 10)

That is, in order to detect the stable state of the comparison processing device, the residual score Q ₀ of the residual matrix E is used for the reference processing device.
Is obtained in advance. Then, constants such as the residual score Q ₀ and the eigenvector obtained by the reference processor are set in the principal component analysis program of the arbitrary processor, and the residual data is obtained from the electrical data of the arbitrary processor under the set conditions. Find the score Q. Then, obtain the difference (deviation amount) from residual scores Q ₀ of the reference processor of residual scores Q of the comparison processing unit, in the comparison processing unit based on the difference between the residual score (Q-Q ₀₎ High frequency power supply 4
It is determined whether the application state of No. 9 has reached a stable state. That is, if the difference (Q−Q ₀ ) in the residual score is large, it indicates that the deviation from the reference processing device is large and unstable, and if the difference (Q−Q ₀ ) is small, the reference _value is small. This indicates that the deviation from the processing apparatus is small and close to a stable state. If the residual score Q ₀ of the reference processor is set to 0, the residual score Q becomes the amount of deviation from the reference level as it is. It is assumed that the values of the variables are calculated so that the average value becomes zero.

The method of operating the processing apparatus according to the present embodiment is performed by appropriately combining the following conditions A and B and processing conditions A and B as shown in (1) to (4) to process the wafer. Basic and its harmonic measurement values V (f _n ), I (f _n ), P
(f _n ), the relative measurement value of Z (f _n ) and the residual score Q are shown in FIG.
As shown in FIG. The principal component analysis result obtained by the reference processing device is set in advance in the principal component program of an arbitrary processing device. The plot in each figure shows the average value per wafer. In the following processing conditions, the value of the deposition is set to 1 when the deposition amount is small,
The condition with a large amount of deposition is shown as a relative value to the condition with a small amount of deposition. I. State A: The processing chamber is evacuated for 12 hours. State B: The processing chamber is evacuated for 4 days. II. Processing conditions Processing conditions A: Conditions with little deposition Wafer processing time: 1 minute High-frequency power: 1700 W Processing chamber pressure: 45 mTorr Process gas: C ₄ F ₈ = 10 sccm, CO = 50 sccm
Ar = 200 sccm O ₂ = 5 sccm Deposition: 1 (relative value) Processing condition B: Conditions with many deposition Wafer processing time: 1 minute High frequency power: 1500 W Processing chamber pressure: 53 mTorr Process gas: C ₄ F ₈ = 16 sccm, CO = 300 sccm Ar = 400 sccm Deposition: 1.95 (relative value)

First, a difference in stabilization due to a difference in processing conditions after maintenance and inspection will be described with reference to FIGS. (1) After the inside of the processing chamber 11 was brought to the state A, the processing apparatus was set to the processing condition A with little deposition. In this state, the wafer was loaded into the processing chamber 11 and processed. Immediately after the wafer is loaded (immediately after start-up), the voltage, current, phase and impedance of the fundamental wave and harmonics of the high-frequency power supply 19 are measured every 0.2 seconds using the harmonic measuring device 17, and the measured values V ( f _n ), I (f _n ), P (f _n ), Z
The average value of (f _n ) for each wafer was determined, and the ratio to the respective values of the reference processing apparatus was taken. FIG. 2 (a) shows how the measured values fluctuated. According to the results shown in FIG. 2A, immediately after the start of the processing, each measured value slowly decreases to the reference value (=
It is determined that the convergence has been achieved in 1) and the reference value level has been substantially reached from the vicinity of the mark, and the state has been stabilized. However, even after the mark, up and down swings are recognized. Also in FIG. 2A, it is easier to determine a stable state compared to the conventional method shown in FIG. On the other hand, according to the method of the present embodiment, the residual score Q
As a result, was obtained as shown in FIG. FIG.
In (a), a plurality of measured values are combined into one as a residual score Q, and the deviation from the reference value can be easily determined even when compared with (a) in FIG. ~ 12
It can be determined that there is no image. Even thereafter, the residual score Q tends to increase slightly periodically.

(2) After the inside of the processing chamber 11 was brought into the above-mentioned state A in the same manner as in (1), processing conditions B with many depositions were set unlike in (1). Then, the wafer is placed in the processing chamber 1
After processing the wafer by loading it into the apparatus 1 and obtaining measured values from immediately after the start of the processing apparatus until the applied state of the high-frequency power supply 19 is stabilized, each measured value is set to the reference value in the same manner as in (1). The results are shown in FIG. 2 (b). FIG.
According to the result shown in (b), each measured value goes to a stable state earlier than in the case of (1). Not much different from the case. On the other hand, when the method of this embodiment is examined, the residual score Q converges to the reference value earlier than in the case of (1) and becomes stable, as is clear from the result shown in FIG. It can be seen that it is easy to judge the point of the stable state. If the residual score for determining the stable state using the reference processing device is determined in advance, the stable state of the comparison processing device can be reliably determined.

Next, a difference in stabilization due to a difference in the state of the processing chamber after the maintenance and inspection will be described with reference to FIGS. (3) After the inside of the processing chamber 11 was brought to the state A, the processing apparatus was set to the processing condition A with little deposition. Then, the wafer is carried into the processing chamber 11, the wafer is processed, and measurement values from immediately after the start of the processing apparatus until the application state of the high-frequency power supply 19 is stabilized are obtained. The ratio with the reference value is taken in the same manner as in the above, and the result is shown in FIG.
It was shown to. According to the result shown in FIG. 4A, it is understood that each measured value slowly converges to the reference value, and the stable state is slowly reached. It is determined that a stable state has been reached at around the ○ mark around 120 wafers. However, even after that, there are measured values that fluctuate up and down, and it is understood that the determination of stabilization is difficult. In contrast, when the method of the present embodiment is examined, the residual score Q converges to the reference value, unlike the result of the measurement value of the harmonic measuring device 17 as is apparent from the result shown in the state A of FIG. It takes an unexpected amount of time to complete, and it turns out that a stable state is reached only when the circle mark is reached.

(4) After the inside of the processing chamber 11 is brought to the state B,
As in the case of (3), the processing apparatus was set to the processing condition A with little deposition. Then, the wafer is carried into the processing chamber 11, the wafer is processed, and measurement values from immediately after the start of the processing apparatus until the application state of the high-frequency power supply 19 is stabilized are obtained. In the same manner as in the above, the ratio with the reference value was taken, and the result is shown in FIG. According to the result shown in FIG. 4B, it is understood that each measured value converges to the reference value earlier than in the case of FIG. When the method of this embodiment is examined, FIG.
As shown in the state B, it is clear that the residual score Q reaches the reference value quickly, but within 100 wafers, it fluctuates and 100 or more wafers are completely stable.

As described above, according to the present embodiment,
Harmonic measuring device 17 between electrode 12 and matching circuit 18
And using the harmonic measuring device 17, the measured values V (f _n ), I (I) of the electrical data such as the voltage value, the current value, the phase, and the impedance of the fundamental wave and the harmonic of the stabilized processing device 10 are obtained. (F _n ), P (f _n ), and Z (f _n ) are used to perform a reference principal component analysis in advance to obtain a reference residual score Q ₀ . Immediately after the start, electrical data is measured by the harmonic measuring device 17 and the measured values V (f _n ), I (f _n ), P (f _n ), Z (f
performing principal component analysis for comparison seeking residual scores Q for comparison with _n), maintenance from both Q, the difference in Q ₀ by comparing the comparison residual scores Q and the reference residual scores Q ₀ Subsequent processing device 10
In order to detect the stable state of the high-frequency power supply, even if there is an enormous amount of measured values, these data are combined into a single residual score Q and the reference value is compared with the processing device 10 after maintenance and inspection. Specifically, a stable state in the processing chamber 11 can be objectively and reliably evaluated and determined. Further, according to the present embodiment, not only can the point of time at which a stable state is reached be evaluated and judged, but also the processing conditions such as the evacuation time in the processing chamber 11 can be set to lead to a stable state. Can be evaluated and judged.

Next, an embodiment of a method for detecting an abnormality of a processing apparatus according to the present invention will be described. The abnormality detection method of the processing apparatus of the present embodiment is also common to the operation method of the processing apparatus of the above embodiment in that the residual score Q in the principal component analysis is used. However, in the present embodiment, a normal processing device, that is, a processing device that is correctly assembled in accordance with the specifications without mounting errors of components and the like in the processing chamber 11 and the high-frequency power supply 19 is used as the reference processing device. In the present embodiment, it is needless to say that the electrical data of the fundamental wave and its harmonics are measured when the applied state of the high-frequency power supply 19 has been released from the unstable state and reached the stable state after the start of the processing apparatus.

Therefore, in the present embodiment, similarly to the above embodiment, the voltage, current, phase and impedance of the fundamental wave and its harmonics related to the reference processing device are intermittently measured as electrical data, and the measurement of each frequency is performed. Value V
(f _n ), I (f _n ), P (f _n ) and Z (f _n ) are obtained. Then, regarding the reference processing device, the residual score Q defined by Expression 9
₀ is obtained in advance. Constants such as eigenvectors obtained by the reference processor are set in a principal component analysis program of an arbitrary processor, and a residual score Q is obtained from electrical data of the arbitrary processor under the set conditions. Then, determine the difference between the residual score Q of residual scores Q ₀ and any processor of the reference processing unit (shift amount), any of the processing apparatus based on the difference of the residual score (Q-Q ₀₎ Is determined to be in a stable state. That is, if the difference between the residual score (Q-Q ₀₎ is large, and its optional processing device processing chamber and / or the high frequency power source 1
9 indicates that there is a mounting error or the like of the component No. 9. Difference (Q-
If Q ₀ ) is equal to or less than the allowable value, the processing device is determined to be normal. When the residual score Q represents a value different from the other residual scores, a row indicating a different value of the residual matrix E, for example, the residual score of the ith measurement result is different from the reference residual score Q. If the values are different, the residual component e of e _{i in} the i-th row
By watching _ij , it is possible to determine which variable (measured value) contributes to the deviation of the residual score Q. From this, the cause of the anomaly and variables with large residuals (fundamental wave,
By associating the voltages and currents of the harmonics, the cause of the abnormality can be classified.

FIG. 6 is a graph showing the relationship between the residual score Q and a component mounting error. In FIG. 6, N1 and N2 are residual scores of a normal processing apparatus, state A is a residual score without screws, state C is a residual score without covers, and state D is a part different from state A. Is the residual score when there is no screw, and the state E is the residual score when there is no cover of a part other than the state C, and the state F is
Indicates a residual score when the screw is loose, and state G indicates a residual score when there is no component. For example, looking at the residual component of the row indicating the residual score in state A, the result is as shown in FIG. If there is no screw in this state A, it can be seen that the voltage and impedance of the fundamental wave swing particularly largely to the minus side, and the current of the third harmonic and the phase of the fundamental wave swing particularly greatly to the plus side. If there is no cover in the state C, it can be seen that the voltage and impedance of the fundamental wave greatly swing to the minus side and the phase relatively swings to the plus side as shown in FIG. If there are no components in the state G, it can be seen that the current and the phase of the fundamental wave greatly swing to the minus side, and the impedance of the fundamental wave swings particularly greatly to the plus side. Therefore, the type of parts,
It is possible to classify the relationship between the attachment site and the like and the component with the largest residual score. By grasping this relationship in advance, it is possible to determine what abnormality is present by knowing a component having a high contribution rate to the residual score.

As described above, according to this embodiment,
After performing principal component analysis in advance using measurement data of the high-frequency power supply 19 of the normal processing apparatus to obtain a residual score for reference,
Principal component analysis is performed using a plurality of measurement data obtained by measuring a plurality of electrical data of an arbitrary processing device to obtain a residual score for comparison. Then, a comparative residual score Q and a reference residual both Q by comparing the score Q _0, since that to detect the abnormality of any of the processing unit from the difference between the Q _0, it is possible to reliably detect the abnormality by the component mounting mistakes without opening the processing apparatus, moreover Mounting errors can be classified from the components of the residual matrix E.

In the above embodiment, the method of operating the processing apparatus and the method of detecting an abnormality using the principal component analysis have been described. However, the present invention can be realized by using a multivariate analysis technique.

[0037]

According to the first to fifth aspects of the present invention, it is possible to objectively determine the stable state of the processing apparatus after starting, and to optimize the conditions for operation. A method of operating the device can be provided.

Further, according to the present invention, an abnormality due to a component mounting error can be reliably detected without opening the processing device, and the mounting error is classified. It is possible to provide a method of detecting an abnormality of a processing device that can perform the processing.

FIG. 1 is a configuration diagram illustrating an example of a processing apparatus to which a method for operating a processing apparatus and an abnormality detection method according to the present invention are applied.

FIGS. 2 (a) and 2 (b) are graphs each showing a transition until electrical data of a processing device is stabilized using a harmonic measuring device.

FIGS. 3 (a) and 3 (b) are FIGS. 2 (a) and 2 (b), respectively.
It is a graph which shows transition until the residual score of electrical data corresponding to (b) is stabilized.

FIGS. 4 (a) and 4 (b) are graphs showing transitions until electrical data of a processing device is stabilized using a harmonic measuring device.

FIGS. 5 (a) and 5 (b) are (a) and (b) of FIG. 2, respectively.
It is a graph which shows transition until the residual score of electrical data corresponding to (b) is stabilized.

FIG. 6 is a graph showing residual scores based on electrical data of a normal processing device and an abnormal processing device.

FIGS. 7A to 7C are graphs each showing a residual component of electrical data of an abnormal processing device.

FIG. 8 is a graph showing a change in electrical data immediately after the start of a conventionally used processing apparatus.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Processing apparatus 11 Processing chamber 12 Electrode 19 High frequency power supply W Wafer (object to be processed)

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H05H 1/46 H01L 21/302 C F term (Reference) 4G075 AA24 AA30 AA41 AA51 AA61 AA65 BC06 CA14 CA25 CA43 CA62 CA65 DA04 DA11 DA18 EB01 EB42 EC21 ED08 FB02 FC11 4K030 FA03 HA11 JA14 JA17 JA18 KA02 KA39 KA41 5F004 AA01 AA09 BA08 BB11 BB13 BD04 CA02 CA03 CA08 CB07 DA00 DA23 DA26 5F045 AA08 BB02 BB03 E03GB10E20 GB20

Claims (10)

[Claims] When a high frequency power is applied to an electrode in a processing chamber from a high frequency power supply to generate a plasma and process an object to be processed, a plurality of electric powers of the high frequency power supply change according to a state in the processing chamber. A multivariate analysis using a plurality of measurement data measured by a measuring device to measure the static data, to detect the application state of the high-frequency power supply, and to operate the processing apparatus. A step of performing a multivariate analysis for reference in advance using the measurement data when the application state of the high-frequency power supply is stabilized according to a state, and measuring a plurality of electrical data of an arbitrary processing apparatus immediately after starting. The step and the step of performing a multivariate analysis for comparison using a plurality of measurement data, and comparing the comparison multivariate analysis result and the reference multivariate analysis result, the high-frequency power source of the arbitrary processing apparatus The above processing chamber Detecting that a stable state has been reached according to the state of the inside of the processing apparatus. 2. When a high frequency power is applied to an electrode in a processing chamber from a high frequency power supply to generate a plasma and process an object to be processed, a plurality of electric powers of the high frequency power supply change according to a state in the processing chamber. A main component analysis using a plurality of measurement data measured by a measuring device that measures the dynamic data, detects the application state of the high-frequency power supply, and operates the processing device, and the method includes: Performing a principal component analysis for reference in advance using the measurement data when the application state of the high-frequency power supply is stabilized according to a state, and measuring a plurality of electrical data of an arbitrary processing apparatus immediately after startup The step of performing a principal component analysis for comparison using a plurality of measurement data; and comparing the result of the comparison principal component analysis with the result of the reference principal component analysis. In the processing chamber Detecting that a stable state has been reached according to the state. 3. The method according to claim 1, wherein at least a voltage value, a current value, an impedance, and a phase angle of each of a fundamental wave and a harmonic are used as the electrical data. . 4. The operation of the processing apparatus according to claim 2, wherein after the residual score or the principal component score is obtained by the principal component analysis, the residual score or the principal component score is compared. Method. 5. The operating method according to claim 2, wherein the processing condition and / or the operating condition are determined based on a result of the comparison of the residual scores. 6. When a high-frequency power is applied from a high-frequency power source to an electrode in a processing chamber to generate a plasma and process an object to be processed, a plurality of electric powers of the high-frequency power source change according to a state in the processing chamber. A multi-variate analysis using a plurality of measurement data measured by a measuring device that measures the dynamic data, detects the application state of the high-frequency power supply, and detects an abnormality of the processing device, the normal processing device Performing a multivariate analysis for reference in advance using the measurement data when the application state of the high-frequency power supply is stabilized according to the state in the processing chamber, and measuring a plurality of electrical data of an arbitrary processing apparatus And the step of performing a multivariate analysis for comparison using a plurality of measurement data, and comparing the comparison multivariate analysis result and the reference multivariate analysis result to detect an abnormality of the above-described arbitrary processing apparatus from a difference between the two. Having a process of A method for detecting an abnormality of a processing device. 7. When a high-frequency power is applied to an electrode in a processing chamber from a high-frequency power source to generate a plasma and process an object to be processed, a plurality of electric powers of the high-frequency power source change according to a state in the processing chamber. A main component analysis using a plurality of measurement data measured by a measuring device for measuring dynamic data to detect an applied state of the high-frequency power supply, thereby detecting an abnormality of the processing device. Performing a principal component analysis for reference in advance using the measurement data when the application state of the high-frequency power supply is stabilized according to the state in the processing chamber, and measuring a plurality of electrical data of an arbitrary processing apparatus Performing a principal component analysis for comparison using a plurality of measurement data; and comparing the comparison principal component analysis result and the reference principal component analysis result to detect an abnormality of the above-described arbitrary processing apparatus from a difference between the two. Having a process of A method for detecting an abnormality of a processing device. 8. The abnormality detection of a processing apparatus according to claim 6, wherein at least a voltage value, a current value, an impedance, and a phase angle of each of a fundamental wave and a harmonic are used as the electrical data. Method. 9. The method according to claim 7, wherein the residual score or the principal component score is obtained by the principal component analysis, and then the residual score or the principal component score is compared.
3. The method for detecting an abnormality of a processing device according to claim 1. 10. The abnormality detection method for a processing device according to claim 6, wherein the abnormal part is classified according to the components of the residual matrix.